Valente, Thomas W. PhD; Fosados, Raquel MPH
PUBLIC HEALTH AGENCIES HAVE BEEN engaged in promoting a wide variety of behaviors, including hygiene, injury prevention, substance abuse prevention, sexually transmitted disease (STD)/HIV prevention, adoption of healthy lifestyles and eating habits, family planning, and contraceptive use, among many others. Implemented in many forms, these promotions have been accompanied by considerable research about their planning and effectiveness. Although many programs in the past have used one medium, television or radio, for example, experts today increasingly agree that multimedia approaches are necessary for the effective promotion of behavior change. Unfortunately, interpersonal communication is one medium that is often omitted from the strategic menu. This article discusses how interpersonal communication can be used and incorporated into health promotion efforts designed to encourage behaviors that reduce the spread of STDs/HIV. The foundation for this article is based on the diffusion of innovations theory, and we begin by discussing its principles followed by an explanation of its use in evaluating mass media campaigns. Throughout, we highlight the importance of interpersonal communication and social networks, concluding with a discussion of social network interventions to promote health.
Behavior Change and the Diffusion of Innovations
Our starting point is the research conducted on how new ideas and practices spread within and between populations using the diffusion of innovations theory.1 Diffusion of innovations theory describes the mechanism in which new ideas, opinions, attitudes, and behaviors spread throughout a community.1–6 Defined by Rogers as “the process by which an innovation is communicated through certain channels over time among the members of a social system,”7,p.5 diffusion theory has been applied to study the spread of new computer technology, educational curricula, farming practices, family planning methods, medical technology, and several other innovations. Diffusion of innovations is a specific incarnation of a hierarchy model,8,9 and the principles of the hierarchy approach have been expanded10 and adapted specifically to the case of family planning.1,11–14
Diffusion theory comprises 5 major components: 1) people pass through stages in the adoption process, 2) diffusion takes time, often a long time, 3) people can modify the innovation and sometimes discontinue its use, 4) perceived characteristics of the innovation influence adoption, and 5) individual characteristics influence adoption. The first 2 components, the time it takes for diffusion to spread and the stages of adoption, are discussed at length.
Diffusion researchers postulate 5 stages in the adoption (behavior change) process: knowledge, persuasion, decision, trial, and adoption.1 A composite index score indicating the stage of adoption can become a convenient outcome variable in the analyses of health promotion programs. Although at first glance, diffusion of innovation's stages of adoption may appear similar to Prochaska's stages of change, the adoption stages indicate a person's progress toward adopting a healthy behavior, whereas stages of change indicate a person's progress toward quitting a deleterious one.15 Diffusion stages tend to emphasize the role of information and sources of influence, whereas stages of change emphasize cognitive dispositions (see reference 14, p. 42 for a comparison of stage models).
Because people become aware of new behaviors at different times, and because they pass through the stages of adoption at different rates, there is considerable time between the first and last adopters of a new behavior. For example, Ryan and Gross showed that despite the innovation having been far superior to the one it replaced, it took 14 years from the first to the last adopter to adopt the use of hybrid seed corn in 2 Iowa counties.3 As far as HIV, the rate of adoption of prevention behaviors has varied. Singhal and Rogers reported that during the course of their HIV intervention program among gay men in San Francisco, the adoption of safer sex behaviors by early adopters occurred within a few years of the start of the intervention. This fast adoption was attributed to the fact that safer sex information was diffused through personal networks.16 On the other hand, in many developing countries, the diffusion and adoption of the HIV ABC model (abstinence, being faithful, or condom use) has been slow.17
This spread of new ideas and practices can be graphed; when done so, it resembles a typical growth or S-shaped curve (Fig. 1). Because diffusion often occurs through personal networks, and personal networks are shaped by geographic, ethnic, and socioeconomic (SES) factors, innovation diffusion tends to be contoured by these factors as well. Thus, there may be different diffusion trajectories for different subgroups (e.g., fast diffusion for high SES segments and slow diffusion for low ones).
Early in the diffusion of a new behavior, there are few adopters and the growth in new adopters is slow. Research has found that these early adopters are often more persuaded by mass media and other targeted media that provide information relevant to the behavior. It is also suspected that these new adopters are sometimes free from social norms that would otherwise inhibit them from adopting a new behavior. Whatever the case, new behaviors are uncertain and risky; these early adopters have to perceive some compensation or reason to adopt the new behavior.
These 2 components, stages of adoption and the time needed for diffusion to occur, are graphed in Figure 1. Depicted here are projected rates of the spread of awareness (knowledge), positive attitude, and behavior (practice). Expected levels for each can be determined by looking at any point in time, and expected lag between awareness and use can be attained by reading across it. This general model, in which awareness (K) eventually leads to a positive attitude (A), which in turn leads to use (P), has been termed the learning hierarchy. Although this sequence (K-A-P) occurs often, others18,19 have argued that different sequences are possible. For example, some behaviors may be adopted first followed by positive attitudes or knowledge of the benefits. For example, many people may use condoms because they want to be safe (practice) although they do not like them (low attitude) and with little knowledge about their effectiveness (knowledge), thus engaging in a P-A-K behavioral change sequence.
Regardless of the behavior change sequence, health promoters have tried to accelerate behavior change by informing the public (changing knowledge), improving attitudes, and directly encouraging individuals to adopt healthy behaviors. Two specific functions of media campaigns have been to: 1) spread knowledge rapidly so that the knowledge curve grows quickly and 2) shorten the KAP gap, the time between awareness and use. Some argue that media campaigns are most effective early in the diffusion process. Because there are few other adopters, individuals seek these adopters out to get information. However, later in the diffusion process, media campaigns serve to get the behavior back on the public agenda16,20 and stimulate further interpersonal communication.21
Mass Media Campaigns
Early studies on the effects of communication campaigns provided results of both successes22–24 and failures.25–27 Many of these studies were conducted by Lazarsfeld and colleagues.28–32 However, a theoretical outcome of this cumulative work was the classic 2-step flow model.33–35 The 2-step flow model posits that opinion leaders use mass media for information more than opinion followers do and these leaders pass on their opinions to these followers. However, since the mid-1950s, no other theory has emerged that integrates mass and interpersonal communication processes within the context of campaign effects, resulting in a balkanized communication field today.36–40
Many scholars have argued that mass media are effective at disseminating information but that interpersonal communication is necessary for behavior change.13,37–44 This adage has directed many projects to use mass media to advertise new ideas and products and then to rely on outreach and peer education programs for adoption.41,42 Few studies, however, have tested the relative influences of mass and interpersonal communication within a particular study.13,43,44 Consequently, there are few models that integrate mass and interpersonal communication influences. This lack of integration helps sustain an underappreciation of the role of mass media in creating sustainable behavior change and has led to an underappreciation of the linkage between mass and interpersonal communication.
Interpersonal Communication Networks
Although we tend to think of media campaigns as broadcasts to a population of individuals, the audience is a web of human relations connected to one another in complex and nonrandom ways. As a result, campaign messages are not received in a vacuum, but rather are filtered through social networks. People often receive messages with others, and this may directly influence the interpretation of the message. Furthermore, people usually speak with others regarding the health promotion messages, thereby causing them to reinterpret it. Consistently, campaigns are designed with the goal of generating interpersonal discussion on the topic in an attempt to set the public's agenda. Such was the case of the STOP AIDS program. Although initially the creators were unaware of the strong educational impact created by the focus groups, it was soon discovered that the focus groups served to generate interpersonal discussions about HIV prevention not only among focus group attendees, but with their personal networks as well. Coupled with media campaigns designed to increase knowledge and awareness of HIV/AIDS among the gay community in San Francisco, the spread of interpersonal discussions generated by this program helped to curtail new infections by the mid-1980s.16,45
As mentioned here, the first attempt to formulate a model linking mass and interpersonal communication was the 2-step flow hypothesis by Katz and Lazarsfeld.29,34 The 2-step flow hypothesis posited that the media influenced opinion leaders who in turn influenced others who are less attentive to media communications. Opinion leaders were found to receive more media and were more aware of current events than no leaders. To persuade others to follow their opinions, opinion leaders used media communications to buttress their arguments. Gladwell writes “These mavens make extensive use of the media to stay expert on their favorite subjects and become trusted sources of information for others.”46
Figure 2 shows a simplified model of the mass media influencing opinion leaders, who in turn influence others. Usually, these others are thought to be family, friends, coworkers, and perhaps acquaintances—generally being people with whom they are close to and have strong credibility and trust with. This model, however, may be an oversimplification. It may be that media influence opinion leaders who influence others that influence others—a 3-step or even multistep flow. Furthermore, it may be that some opinion leaders influence one or a few others, whereas others have much higher multiplier effects, influencing 5, 10, or hundreds of others. Figure 3 attempts to capture this complexity.
These models, however, neglect to consider a number of other factors regarding the media influence process. First, it is likely that opinion leaders are influenced by others as much as others are influenced by them and that media shape their messages in accordance with what they think the audience wants to see or hear. In summary, to say that media communications can influence A whom influences B may be an oversimplification.
Second, individuals are embedded within complex social network structures. Some people have small networks, whereas others have quite large ones. Some social networks are integrated (their friends know each other), whereas others are radial (their friends do not know one another). What follows are 3 ways in which social network structures can affect media processes. First, Potterat, Rothenberg, and Muth proposed that network structure, especially the cohesiveness of an individual's network, is associated with STD/HIV risk for the individual.47,48 The authors report that lower cohesiveness of network members is associated with lower STD/HIV transmission, even in a high-risk population. Second, the norms held within social networks alter the media influence process. For example, if a social network comprised of young adults has negative safer sex norms, a media campaign designed to target and change these norms may increase condom use.49 Third, the degree of similarity or difference between a person and his or her social contacts (homophily) also affects the flow of ideas and behaviors. It is believed that information flows and persuasion occurs more readily among homophilous dyads, that is, people who are like one another, rather than among nonhomophilous dyads.1 Consequently, diffusion, as mentioned earlier, tends to occur along sociodemographic lines because social networks are contoured by sociodemographic characteristics.
Finally, there is variation in risk taking and risk avoidance in the population. This can also affect the media influence process. It is well known that there is considerable variation in the amount of influence required for a person to adopt a behavior. Some people will adopt new behaviors when only a minority of their friends or the population has done so. Others wait until a practice is widely accepted before they are willing to adopt it.5
These factors show that the relationship between mass media and interpersonal communication is complex. Unlike the simple opinion leader models presented here, it is more likely that people attend to media communication, and then interpret and discuss it in unanticipated ways. For example, an antitobacco campaign may be parodied by the intended audience, resulting in boomerang (opposite) effects rather than antitobacco effects. This boomerang effect has also occurred with HIV campaigns. Some members of the gay community have begun to refuse participation in safer sex behaviors because of the widespread access to antiretroviral drugs, believing the risk for HIV to be diminished.17 In other words, the effect of media communications on individuals is a function of how the messages are interpreted within the context of people's social networks—how, with whom, and in what ways the messages are discussed.
Given this reality, researchers should include in measures of program exposure questions that ask who the subject watched the program with or who they participated in the activities with. Researchers should then ask respondents whether these others approved of the messages or activities and, importantly, whether they discussed the messages or activities. Then ask, if they did discuss them, what did each say to each other. Information about who they discussed messages with and what was said could provide valuable information for understanding program effects and designing future activities.
Selecting a Channel
Selecting the appropriate channel, regardless of message, can profoundly influence the effectiveness of behavior change programs. Sometimes, people want and attend to impersonal media, whereas at other times, they prefer personal media. The degree to which programs use impersonal and personal media may directly correlate with program effectiveness. Keeping in mind the interplay between mass and interpersonal communication during the behavior change process, we now turn to a consideration of the channels used to disseminate information or otherwise persuade a group to engage in healthy behaviors.
Table 1 describes 6 ecologic levels that provide the context and reference for health promotion interventions. At each level, different types of promotions, each with distinct advantages and disadvantages, can be made. For example, the individual level can be targeted at clinics and treatment sites with receptionists and providers trained to deliver health promotion materials. Individuals visiting these sites can receive a variety of safer sex materials designed to inform them and help them make healthier choices. A limitation to individual-level interventions is the fact that many people in need of health promotion and treatments do not visit clinics or other treatment centers. As predicted by Kegeles, Hays, and Coates,42 men who reported high risk-taking sexual activities (e.g., unprotected sex, multiple sex partners) with other men were less likely to attend small group meetings, less likely to participate in the Mpowerment Project's formal outreach team, and even less likely to be members of the core group, which was the project's decision-making group.
Organizational interventions are designed to reach people in their schools, worksites, and in the community organizations to which they belong. These interventions have the advantage of reaching people in places where they naturally congregate. It also provides a captive audience for promotional programs. More often than not, organizations are willing and able partners to such efforts. In fact, community-level programs explicitly partner with community groups to create culturally sensitive and culturally tailored programs designed to reach people “where they live, work, and play.” Such interventions are based on principles of empowerment and community efficacy.50 For STD/HIV prevention programs in the United States, many have turned to churches as a natural community and organizational setting to promote healthy behaviors. Churches have strong bonds with their communities and represent credible and trusted sources of strength and support. A number of programs have used churches for health promotion activities with considerable success.51,52
Mass media interventions can be cost-effective ways to reach a large audience with important information such as the availability of treatment services, information hotlines (800 numbers), and web sites. Increasingly, scholars have found 2 critical elements influencing program effectiveness of mass media interventions. First, the communications should be entertaining. Entertainment appeals to an audience and captures their attention. By entertaining the audience, health promotions can reach people who normally would not attend to health messages and who would normally not be inclined to listen to promotions. Because exposure is a requirement for communication effects regardless of the medium, incorporating entertaining communications will usually increase the number of people who receive the message.
Second, personalize the message. Because interpersonal communication is often a requirement for behavior change, using personal models and testimonials with whom the audience can identify can help increase the effectiveness of communications. These methods also serve to prompt interpersonal communication. Communications, whether personal or impersonal, should be personalized.
Finally, at the highest ecologic level is the creation of policy changes that facilitate behavior change. Although often not thought of as a communication strategy, policy change is often a communication strategy in the following ways. First, lobbying and advocating for policies requires interpersonal communication with lawmakers and policymakers. Second, mass media campaigns are often used to mobilize public opinion to create pressure for policy changes. Finally, policy changes are often accompanied by media campaigns promoting the change so that the recipients of the new policy can benefit.
In summary, there are different channels available at different ecologic levels to strategically plan communication for safer sexual practices and other healthy behaviors. Selection of a particular strategy should be guided not only by practicality, but also as one that allows for the opportunity to combine impersonal and personal communications in appealing and entertaining ways so the audience attends to the message and internalizes it. These 2 factors (entertaining and personalization) increase the likelihood that the messages will be discussed within personal networks, thus facilitating behavior change.
Social Network Segmentation
One characteristic of marketing communication is audience segmentation, the partitioning of audiences into distinct groups to tailor messages specifically for each group. Segmentation comes in many forms: 1) geographic segmentation has messages targeted based on region (using southern accents for the south and northern ones for the north); 2) sociodemographic segmentation occurs when messages are tailored on gender, age, ethnicity, or socioeconomic status; and 3) psychographic segmentation uses psychographic profiles that classify individuals on dimensions such as individualism, adventurousness, risk taking, and so on.
We suggest a fourth type of segmentation: sociometric. Sociometric segmentation occurs when messages are targeted to individuals based on their social network position. For example, messages aimed at opinion leaders, particularly those identified through social network analysis techniques, would be an example of sociometric segmentation. Amirkhanian et al53 illustrate this in an HIV prevention program. Using sociometric measures to identify and train the social network leaders of 14 groups of young men who have sex with men (YMSM; n = 82 total network members), the authors reported a significant increase in condom use with casual partners among network members, improved safer sex social norms (P = 0.001), increased network-level AIDS risk reduction knowledge (P = 0.001), and a 15% increase in comfort level speaking about AIDS with network members at follow up. These changes in safer sex practices occurred after having trained the social network leaders of each network how to “effectively communicate HIV prevention messages and personal risk-reduction advice to members of their networks.”53,p.212
Another example of sociometric segmentation would be to create messages aimed at core groups within sexual networks, defined as “individuals that have large numbers of sexual contacts so as to have the potential to infect large numbers of individuals if they themselves get infected.”54,p.263 Core group members who engaged in unprotected sexual contact can accelerate the spread of STDs/HIV.47,55 In geographic regions where there is a high presence of core transmitters, disease can still spread even when cohesion among network members is low and there is high disease prevalence in the area.56 Laumann and Youm stress the importance of creating interventions for STD/HIV prevention by focusing on this structural aspect of the network.57
These are just a few of the many forms sociometric segmentation can use (just as there are multiple variables used in the other forms of segmentation). In contrast to targeting opinion leaders, one could design messages for social network isolates, individuals with few or no social contacts. Messages that appeal to isolates would likely be different than those that appeal to leaders.
Audiences could also be segmented based on whether a person has many friends who engage and/or support a behavior versus those who have few. For example, it may be that the majority of community members support condom use to prevent HIV, but there are some who resist. Messages targeted to these individuals would stress that many others have already adopted safer sex practices, and there are many social supports available to them if they desire. On the other hand, if people who have few personal contacts are identified because they have tried the innovation, the resulting message might stress that the potential adopter will be “on the cutting edge” or a “leader of the pack” by adopting. This type of segmentation would be on personal network exposure to the behavior.
Social network analysts have devised numerous ways to identify positions in a network and often decompose networks into groups and positions. Any of these procedures, which are numerous, could be used to create interventions. In fact, Aral suggests that behavioral change may be easier to maintain if interventions based on structural characteristics of the network are created and implemented.54 For example, a web tool could be created in which users enter the names of their friends and the friends' risk behaviors. The information could then be visually reported back to the user with an estimate of his or her risk status based on their personal network's risk behavior.
Social network segmentation would also lead to designing interventions that specifically focus on creating communications that generate greater interpersonal communication within these networks and provide people with the tools needed to have these conversations. For example, entertainment education strategies58 may be effective at creating behavior change because it generates interpersonal discussion about the topic among viewers.
Social Networks as Delivery Vehicles
Because peer influence and peer modeling are significant correlates of behavior and behavior change,59,60 harnessing the power of peer influence for health promotion seems logical. This power is most likely to be harnessed by allowing peers to deliver health messages. Although peer leaders are defined in different ways (see 61 for a description of a popular opinion leader model), peers have been used in many settings with generally favorable results.16,41,42,53,60–66 In STD/HIV research, using peer leaders as advocates for safer sex practices has been effective in reducing a number of risky sexual behaviors.41,42,53,62–64
Scholars have recently proposed 2 approaches to improve peer delivery of health messages in ways that capitalize on the local nature of opinion leadership. The group model involves identifying groups within social networks first (these can include subgroups and cliques) and then identifying peer opinion leaders within each group. This model was proposed and pilot-tested in a school-based tobacco prevention program by Wiist and Snyder67 and was further developed and implemented by Buller and colleagues68,69 in a worksite nutrition promotion program. This model starts by identifying subgroups or cliques in a network and then selects peer opinion leaders from within each clique. The peer opinion leaders are identified as those people who received the most nominations in response to the question, “Whom do you go to for advice?” A graphic representation of this model is depicted in Figure 4. Figure 4A displays a social network of advice giving among 45 medical doctors in one Illinois community (data are from70). Figure 4B shows the same network with groups defined and leaders identified within each group.
This model was also applied more recently in an HIV prevention program designed to target intact social networks of YMSM who are at highest risk of becoming infected.53 After using ethnographic techniques to identify 16 separate networks, an “index” member or peer opinion leader of each network was identified and approached. The indices served as points of entry for each social network. Once the remaining social network members were recruited, each member was asked to nominate someone within their network who they most and who they least liked about 10 different domains. The individual with consistently high scores as based on sociometric indicator scores served as their social network's leader.53
The network leader model begins with identifying peer opinion leaders first and then grouping network members to each leader, thereby creating new cliques. This model was proposed by Valente and Davis71 and tested in a school-based smoking prevention program.65 Opinion leaders were defined as those members of a network who receive the most nominations as people others go to for advice. Once the leaders are identified, other network members are matched to each leader based on their “closeness” in the social network. Figure 4C rearranges the network so that the leaders are selected and members assigned to the leaders they nominated or are next closest to. As far as we know, this method has not been used in STD/HIV prevention.
Table 2 reports the ID numbers of leaders and the social network members for the 2 different models described. The 2 approaches yield different leaders (2 of the 3 leaders differ) and different network members. So, although the conceptual approaches of the 2 models are similar, the resulting leader and network member definitions are quite different. It also warrants noting that a number of assumptions are built into the realization of both methods. For the group model, there are numerous ways to identify subgroups/cliques depending on whether the individual subgroups/cliques need to be the same size and whether the subgroups/cliques need to be mutually exclusive. For the networked leader model, leaders were chosen based only on a count of the number of nominations they received, not whether they were received by the same or different people. Both methods can capitalize on the order of nominations made, adding more weight to the first nomination, for example, and both need to make assumptions about how one treats isolates.
Studies using social networks as delivery vehicles have reported considerable success.53,65,71 The approach is both intuitively appealing and buttressed by behavior change theories that stress the importance of opinion leadership, homophily, and group dynamics. The main limitation to these approaches is that they are used primarily in closed settings (schools, workplaces, organizations) but may not be feasible in large community and population settings. In a city of tens of thousands, it is not feasible to identify leaders and assign them to those who nominated them or to identify networks/cliques based on social network ties.
This article set out to use behavior change theory, particularly diffusion of innovations theory, to explore its implications for behavior change programs. We opened by presenting the principles of diffusion of innovations theory, noting the importance of behavior changes sequences. We then discussed the history of media campaigns and how their effects have been conceptualized. Throughout, we stressed the importance of interpersonal communication in campaign effects.
Early models of the link between mass media and interpersonal communication were presented, reviewing the 2-step and multiflow hypothesis. The 2-step flow hypothesis proposed that the media influence opinion leaders who in turn influence their less attentive peers. Support for the 2-step flow hypothesis has not been supplied, primarily because individuals inhabit multiple and complex social networks. Given the complexity of interpersonal dynamics and behavior change process, we discussed interventions in the context of ecologic levels of influence and how different interventions are developed at different levels. Although different ecologic levels warrant different types of interventions, it should be stressed that principles of entertainment and interpersonal communication can and should still inform these interventions.
We then introduced the concept of sociometric segmentation as an extension of geographic, sociodemographic, and psychographic segmenting. Sociometric segmentation proposes to identify audiences based on their position in a social network, whether an opinion leader, isolate, or bridge. Furthermore, it can be used to tailor messages to those with many adopters in their network or those with few. Again, we stress that interventions could also be used to generate interpersonal communication in these social networks.
A natural outgrowth of the focus on social networks is to use them as delivery vehicles. The messenger is as important as the message. Consequently, methods are being developed that use naturally occurring social network structures to deliver health promotion programming. Two similar models were reviewed, a group model and a networked leader one. In the group model, subgroups/cliques in the network are identified and leaders chosen from within the group. In the networked leader model, opinion leaders are identified and then members assigned to leaders they choose or are “nearest” in the network. Data from a classic diffusion study are used to illustrate.
We have traced a historic path that started with early conceptualization of how mediated communications can be used to change behavior. The path has been lit with a focus on interpersonal communication, which led naturally to conceiving behavior change processes within social networks. We have proposed that scholars use these social networks as delivery vehicles, channels all too often ignored in our attempts to spread messages to a large audience. The realization that who delivers the message, and in what interpersonal context, may be just as if not more important than the message itself, and may result in better, more relevant, and perhaps more effective programs. All mass media are personal media and used in interpersonal ways.
Clearly, it is too early to judge whether network-based interventions provide a significant benefit to STD/HIV prevention or treatment programs. Initial evidence is promising, and continued application is warranted both to expand the approaches outlined here and to document effectiveness. Future studies should evaluate the added benefit of network information for determining who should deliver messages and in what interpersonal and group settings. Future studies should also measure interpersonal communication about prevention messages to determine how people receive these messages and how their networks moderate or mediate program effectiveness.
1.Rogers EM. Diffusion of Innovations, 5th ed. New York: Free Press, 2003.
2.Katz E, Levine ML, Hamilton H. Traditions of research on the diffusion of innovation. Am Sociol Rev 1963; 28:237–253.
3.Ryan B, Gross NC. The diffusion of hybrid seed corn in two Iowa communities. Rural Sociol 1943; 8:15–24.
4.Valente TW. Diffusion of innovations and policy decision-making. J Commun 1993; 43:30–45.
5.Valente TW. Network Models of the Diffusion of Innovations. Cresskill, NJ: Hampton Press, 1995.
6.Valente TW, Rogers EM. The origins and development of the diffusion of innovations paradigm as an example of scientific growth. Sci Commun 1995; 16:238–269.
7.Rogers EM. Diffusion of Innovations, 4th ed. New York: The Free Press, 1995.
8.Ray ML. Marketing communication and the hierarchy-of-effects. In: Clarke P, ed. New Models for Mass Communication Research. Newbury Park, CA: Sage, 1975.
9.Thorson E. Processing television commercials. In: Dervin B, Grossberg L, O'Keefe BJ, Wartella E, eds. Rethinking Communication, Vol. 2: Paradigm Exemplars. Newbury Park, CA: Sage, 1989:397–410.
10.Mcguire WJ. Theoretical foundations of campaigns. In: Rice RE, Atkin CK, eds. Public Communication Campaigns. Newbury Park, CA: Sage.
11.Rogers EM. Communication Strategies for Family Planning. New York: Free Press, 1973.
12.Piotrow PT, Kincaid DL, Rimon JG, Rinehart W. Health Communication: Lessons From Family Planning and Reproductive Health. Westport: Praeger, 1997.
13.Valente TW, Poppe PR, Merritt AP. Mass media enerated interpersonal communication as sources of information about family planning. J Health Commun 1996; 1:259–273.
14.Valente TW. Evaluating Health Promotion Programs. Oxford: Oxford University Press, 2002.
15.Prochaska JO, DiClemente CC. Toward a comprehensive model of change. In: Miller WR, Neather N, eds. Treating Addictive Behaviors: Processes of Change. New York: Plenum, 1986.
16.Singhal A, Rogers EM. Combating AIDS: Communication Strategies in Action. New Delhi: Sage Publications, 2003.
17.Bertrand JT. Diffusion of innovations and HIV/AIDS. J Health Commun 2004; 9:113–121.
18.Chaffee SH, Roser. Involvement and the consistency of knowledge, attitudes, and behaviors. Commun Res 1986; 13:373–399.
19.Valente TW, Paredes P, Poppe PR. Matching the message to the process: The relative ordering of knowledge, attitudes, and practices in behavior change research. Human Commun Res 1998; 24:366–385.
20.McCombs ME, Shaw DL. The agenda setting function of the press. Public Opin Q 1972; 36:176–187.
21.Wolitski RJ, Bensley L, Corby NH, et al. Sources of AIDS information among low-risk and at-risk populations in five U.S. cities. AIDS Community Demonstration Projects. J Community Health 1996; 21:293–310.
22.Cartwright D. Some principles of mass persuasion: Selected findings of research on the sale of United States War Bonds. Human Relat 1949; 2:253–267.
23.Mendolsohn H. Some reasons why information campaigns can succeed. Public Opin Q 1973; 37:50–61.
24.Rogers EM, Storey JD. Communication Campaigns. In: Berger CR, SH, eds. Chaffee Handbook of Communication Science. Newbury Park, CA: Sage, 1987.
25.Hyman HH, Sheatsley PB. Some reasons why information campaigns fail. Public Opin Q 1947; 11:412–423.
26.NPR National Public Radio Public Service Announcements. Radio Broadcast on the Program ‘All Things Considered.' Washington, DC: National Public Radio, April 16, 1996.
27.Udry JR, Clark LT, Chase CL, Levy M. Can mass media advertising increase contraceptive use? Fam Plann Perspect 1972; 4:37–44.
28.Berelson B, Lazarsfeld PF, McPhee W. Voting: A Study of Opinion Formation in a Presidential Campaign. Chicago: University of Chicago Press, 1954.
29.Katz E, Lazarsfeld P. Personal Influence. New York: The Free Press, 1955.
30.Lazarsfeld PF, Berelson B, Gaudet H. The People's Choice, 2nd ed. New York: Columbia University Press, 1948.
31.Merton RK. Social Theory and Social Structure. Glencoe, IL: Free Press in Maletezke, 1949.
32.Eulau H. The Columbia studies of personal influence. Soc Sci Hist 1980; 4:207–228.
33.Katz E. The two-step flow of communication: An up-to-date report on a hypothesis. Public Opin Q 1957; 21:61–78.
34.Katz E. Communication research since Lazarsfeld. Public Opin Q 1987;S57.
35.Gitlin T. Media sociology: The dominant paradigm. Theory & Society 1978; 6:205–253.
36.Barnett GA, Danowski JA. The structure of communication: A network analysis of the International Communication Association. Human Commun Res 1992; 19:264–285.
37.Chaffee SH. Mass media and interpersonal channels: Competitive, convergent, or complementary? In: Gumpert G, Cathart R, eds. Inter/Media: Interpersonal Communication in a Media World. New York: Oxford University Press, 1982.
38.Hawkins RP, Wiemann JM, Pingree S, eds. Advancing Communication Science: Merging Mass and Interpersonal Processes. Newbury Park, CA: Sage, 1988.
39.Reardon KK, Rogers EM. Interpersonal versus mass media communication: A false dichotomy. Human Commun Res 1988; 15:284–303.
40.Rice RE, Borgman CL, Reeves B. Citation networks of communication journals, 1977–1985: Cliques and positions, citation made and citations received. Human Commun Res 1988; 15:256–283.
41.Kelly JA, Murphy DA, Sikkema KJ, et al. Randomised, controlled, community-level HIV-prevention intervention for sexual-risk behaviour among homosexual men in U.S. cities. Lancet 1997; 350:1500–1505.
42.Kegeles SM, Hays RB, Coates TJ. The Mpowerment project: A community-level HIV prevention intervention for young gay men. Am J Public Health 1996; 86:1129–1136.
43.Hornik RC. The knowledge-behavior gap in public information campaigns: A development communication view. In: Salmon CT, ed. Information Campaigns: Balancing Social Values and Social Change. Newbury Park, CA: Sage Annual Reviews, 1989.
44.Hornik RC. Channel effectiveness in development communication programs. In: Rice RE, Atkin CK, eds. Public Communication Campaigns, 2nd ed. Newbury Park, CA: Sage, 1989.
45.Wohlfeiler D. Community organizing and community building among gay and bisexual men: The STOP AIDS Project. In: Minkler M, ed. Community Organizing and Community Building for Health. 1998:30–243.
46.Gladwell M. The Tipping Point. London: Little Brown and Co, 2003.
47.Potterat JJ, Rothenberg RB, Muth SQ. Network structural dynamics and HIV transmission. Int J STD AIDS 1999; 10:182–185.
48.Rothenberg RB, Potterat JJ, Woodhouse DE, Muth SQ, Darrow WW, Klovdahl AS. Social network dynamics and HIV transmission. AIDS 1998; 12:1529–1536.
49.Friedman SR, Flom PL, Kottiri BJ, et al. Consistent condom use in the heterosexual relationships of young adults who live in a high-HIV-risk neighbourhood and do not use ‘hard drugs.' AIDS Care 2001; 13:285–296.
50.Glanz K, Rimer BK, Lewis FM. Health Behavior and Health Education: Theory, Research, and Practice, 3rd ed. San Francisco: Jossey-Bass Publishers, 2003.
51.Oexmann MJ, Ascanio R, Egan BM. Efficacy of a church-based intervention on cardiovascular risk reduction. Ethn Dis 2001; 11:817–822.
52.Sanders EC. New insights and interventions: Churches uniting to reach the African American community with health information. J Health Care Poor Underserved 1997; 8:373–375.
53.Amirkhanian YA, Kelly JA, Kabakchieva E, McAuliffe TL, Vassileva S. Evaluation of a social network HIV prevention intervention program for young men who have sex with men in Russia and Bulgaria. AIDS Educ Prev 2003; 15:205–220.
54.Aral SO. Sexual network patterns as determinants of STD rates: Paradigm shift in the behavioral epidemiology of STDs made visible. Sex Transm Dis 1999; 26:262–264.
55.Klovdahl AS. Networks and pathogens. Sex Transm Dis 2001; 28:25–28.
56.Ellen JM, Jennings JM, Meyers T, et al. Perceived social cohesion and prevalence of sexually transmitted diseases. Sex Transm Dis 2004; 1:117–122.
57.Laumann EO, Youm Y. Racial/ethnic group differences in prevalence of sexually transmitted diseases in the United States: A network explanation. Sex Transm Dis 1999; 26:250–260.
58.Singhal A, Rogers EM. Entertainment-Education—A Communication Strategy for Social Change. Mahwah, NJ, London: Lawrence Erlbaum Associates, 1999.
59.Alexander C, Piazza M, Mekos D, Valente TW. Peers, schools, and adolescent cigarette smoking: An analysis of the national longitudinal study of adolescent health. J Adolesc Health 2001; 29:22–30.
60.Valente TW, Watkins SC, Jato MN, et al. Social network associations with contraceptive use among Cameroonian women in voluntary associations. Soc Sci Med 1997; 45:677–687.
61.Kelly JA. Popular opinion leaders and HIV prevention peer education: Resolving discrepant findings, and implications for the development of effective community programmes. AIDS Care 2004; 16:139–150.
62.Kelly JA, St. Lawrence J, Stevenson L, et al. Community AIDS/HIV risk reduction: The effects of endorsements by popular people in three cities. Am J Public Health 1992; 82:1483–1489.
63.Sikkema KJ, Kelly JA, Winett RA, et al. Outcomes of a randomized community-level HIV prevention intervention for women living in 18 low-income housing developments. Am J Public Health 2000; 90:57–63.
64.Kelly JJ, St. Lawrence JS, Diaz YE, et al. HIV risk behavior reduction following intervention with key opinion leaders of population: An experimental analysis. Am J Public Health 1991; 81:168–171.
65.Valente TW, Hoffman BR, Ritt-Olson A, et al. Effects of a social-network method for group assignment strategies on peer-led tobacco prevention programs in schools. Am J Public Health 2003; 93:1837–1843.
66.Tobler N. Updated meta-analysis of adolescent drug prevention programs. In: Montoya C, Ringwalt C, Ryan B, Zimmerman R, eds. Evaluating School-Lined Prevention Strategies: Alcohol, Tobacco, and Other Drugs. San Diego: University of California Extension Division, 1993:71–86.
67.Wiist WH, Snider G. Peer education in friendship cliques: Prevention of adolescent smoking. Health Educ Res 1991; 6:101–108.
68.Buller DB, Morrill C, Taren D, et al. Randomized trial testing the effect of peer education at increasing fruit and vegetable intake. J Natl Cancer Inst 1991; 91:1491–1500.
69.Buller D, Buller MK, Larkey L, et al. Implementing a 5-a-day peer health educator program for public sector labor and trades employees. Health Educ Behav 2000; 27:232–240.
70.Coleman JS, Katz E, Menzel H. Medical Innovation: A Diffusion Study. New York: Bobbs Merrill, 1966.
71.Valente TW, Davis RL. Accelerating the diffusion of innovations using opinion leaders. Ann Am Acad Pol Soc Sci 1999; 566:55–67.